A total of 298 children from six treatment unions and 257 children from six control unions were screened; 237 and 227 children, respectively fulfilled the inclusion criteria; 206 mother-child pairs were randomly selected for each group. Loss to follow-up was lower than anticipated. Six treatment children and nine control children dropped out; key reasons included temporary reallocation, migration, and refusal to give data and/or anthropometric measurements every month. One child died of drowning and another died of illness in the treatment area. In the control area, one child also died of drowning (Figure 1).
Background characteristics of the households and mother-child pairs
Table 1 presents the baseline demographic, child and maternal characteristics. The average weight of the children was 8.2 kg and their age (around 9 months) was similar between the groups at baseline. Similar proportions of children were male and female. The treatment group had a higher average length (69.1±3.0 cm) than the control group (68.4±3.1 cm). However, mean LAZ did not differ between groups (treatment: -1.01±0.89; control: -1.07±0.95). The majority of the children were breastfed. Maternal mean age was similar between groups (treatment: 25.1±4.9, control: 24.9±5.4, years); around 56% of mothers in both groups had secondary or higher levels of education and equal proportions (6.3%) of mothers in both groups were illiterate. Access to safe drinking water and a sanitary latrine was comparable between groups. A higher proportion of households were food secure in the control group than treatment group (control: 67.0% and treatment: 42.7%).
Effect of intervention on child growth at different follow up
Table 2 presents the primary outcome analysis. The treatment group achieved a significantly greater length gain than the control group in the 8th month (78.79 and 77.17 cm; p<0.001) with a difference in length of 1.62 cm between the groups; this difference increased to 1.99 cm (81.14 and 79.15 cm; p<0.001) and 2.62 cm (83.52 and 80.89 cm; p<0.001) respectively at 10th and 12th months. The mean LAZ score was significantly higher in the treatment group than control group from 10 months onwards. By the end of month 12, the mean LAZ scores of the treatment and control groups were -0.62 and -1.16 (p<0.001). The GLM regression model adjusted for children’s age, sex, baseline anthropometric measurements, maternal height and household asset index. The IRR of LAZ score was 0.38 higher in the treatment group (CI: 0.24, 0.51, p<0.001) at endpoint. Risk of stunting was 61% lower in the treatment group (IRR: 0.39, CI: 0.22, 0.67, p<0.001) at 12th month.
Effects of intervention on dietary intake, handwashing and sanitation practices
We collected dietary data every month using previous-day 24-hour dietary recall (Table 3). The treatment group children consumed significantly more meat, fish or fleshly food than the control group. Although the egg-/milk-based snack was the major intervention component, we also assessed the intake of eggs and milk not included in the snack. Over 95% of treatment group children had consumed eggs in the previous 24 hours at 4 and 8-month follow-up; egg intake slightly decreased to 91% in the intervention group at 12th month. Minimum dietary diversity and minimum acceptable diet was 25% and 28% higher in the treatment group than in the control at 12th month respectively. Hand-washing and sanitation practice data were collected every two months and composite scores were developed for household hand-washing practices and facilities. Hand washing practice scores differed significantly between the treatment and control groups at 8th and 12th month; the facility score only differed significantly at 8th month (Table 3).
Child morbidities during the intervention
Morbidity data (14-day recall) was collected monthly; cumulative morbidity is presented in Table 4. Total fever episodes over the 12-month intervention were significantly lower in the control than treatment group (cof: 0.53, CI: 0.19, 0.89) after adjusting for age and sex. Cumulative duration of fever was 1.62 days longer in the treatment group (CI: 0.03, 3.21). Cumulative number of episodes of diarrheal illness did not differ between groups; however, the cumulative duration of diarrhea and dysentery were significantly lower in the treatment group (cof: -1.17, CI: -2.27, -0.08). Numbers of episodes and duration of coughs, colds and runny noses were comparable between the two groups.